Article pubs.acs.org/est
Understanding Receptor-Mediated Effects in Rainbow Trout: In Vitro−in Vivo Extrapolation Using Physiologically Based Toxicokinetic Models Markus Brinkmann,*,† Kathrin Eichbaum,† Sebastian Buchinger,‡ Georg Reifferscheid,‡ Thuy Bui,§ Andreas Schaff̈ er,§,∥ Henner Hollert,†,∥,⊥,# and Thomas G. Preuss§ †
Department of Ecosystem Analysis, RWTH Aachen University, Institute for Environmental Research, Worringerweg 1, 52074 Aachen, North Rhine-Westphalia, Germany ‡ Department G3: Biochemistry, Ecotoxicology, Federal Institute of Hydrology (BfG), Am Mainzer Tor 1, 56068 Koblenz, Rhineland-Palatinate, Germany § Chair of Environmental Biology and Chemodynamics, RWTH Aachen University, Institute for Environmental Research, Worringerweg 1, 52074 Aachen, North Rhine-Westphalia, Germany ∥ State Key Laboratory of Pollution Control and Resource Reuse, School of the Environment, Nanjing University, Nanjing, Jiangsu 210093, China ⊥ College of Resources and Environmental Science, Chongqing University, Chongqing 400044, China # College of Environmental Science and Engineering and State Key Laboratory of Pollution Control and Resource Reuse, Tongji University, Shanghai 200092, China S Supporting Information *
ABSTRACT: The European REACH regulation requires the use of animal experimentation to assess the risk of industrial chemicals. However, the 3R principle (reduction, replacement, refinement) demands the use of suitable alternative test methods. Many dossiers submitted for the authorization of chemicals have attempted to provide the required data without performing new experiments, relying heavily on in silico methods; in vitro assays were scarcely used. We propose a methodology that uses physiologically based toxicokinetic (PBTK) models to extrapolate in vitro data to the in vivo level. We collected experimental results for in vitro and in vivo ethoxyresorufin-O-deethylase and vitellogenin induction following chemical exposure and compared those results with model predictions. We found that the predictive power of aqueous chemical concentrations was limited; median effect concentrations (EC50s) based on internal concentrations in fish correlated better with in vitro EC50s. Our data show that in vitro assays could offer a substitute for fish studies when combined with PBTK models.
■
INTRODUCTION The Industrial Revolution of the 19th century marked the beginning of the Anthropocene, an era in which human activities have triggered global environmental changes.1 Such human activities have placed pressure on important planetary boundaries, i.e. biophysical thresholds that, when exceeded, could have disastrous consequences for humanity.2 Chemical pollution is one such boundary that remains undefined. In many developed countries, regulations of varying rigor have been established to control chemical pollution. For example, regulation No. 1907/2006, which concerns the registration, evaluation, authorization, and restriction of chemicals (REACH) was established by the European Parliament and the Council to prospectively avoid the negative effects of industrial chemicals on humans and the environment.3−5 © 2014 American Chemical Society
To meet this mandate, the chemical industry must provide toxicological and ecotoxicological data, potentially requiring a large number of animal experiments.6 In a 2011 report following the REACH program’s first registration period, however, the European Chemicals Agency (ECHA) reported that data from new animal experiments were included in less than 1% of the dossiers for all assessed toxicological effects,7,8 and many dossiers that did not perform new animal experiments marginally provided the required information. In most cases, the data were generated using in silico methods (e.g., quantitative Received: Revised: Accepted: Published: 3303
December 2, 2013 February 16, 2014 February 18, 2014 February 18, 2014 dx.doi.org/10.1021/es4053208 | Environ. Sci. Technol. 2014, 48, 3303−3309
Environmental Science & Technology
Article
mechanism-specific end points (i.e., those caused by chemicals with specific modes of toxic action) compared to acute effects. In the present study, we attempted to overcome this limitation by using a dynamic, multicompartment, physiologically based toxicokinetic (PBTK) model for the bioconcentration of organic chemicals in rainbow trout (Oncorhynchus mykiss).18,19 This type of model is capable of predicting a chemical’s internal concentration in the whole fish and in various tissues at any time postexposure, making it useful for quantitative in vitro−in vivo extrapolation (IVIVE) in fish.20,21 The PBTK model also facilitates a link between the biochemical effect (e.g., EROD activity), which is measured in an animal’s liver, to the chemical concentration at the site of toxic action. We collected a comprehensive data set for two receptor-mediated modes of action (MOA), namely EROD and Vitellogenin (Vtg) induction, in rainbow trout measured (a) in vivo and (b) in vitro using primary hepatocytes. Vtg is a biomarker for endocrine disruption in fish, a central MOA within REACH.3,22 EROD is a common exposure biomarker for persistent contaminants (e.g., dioxins).23 Using EC50 values measured in vivo, the corresponding internal concentrations in whole animals and their livers were calculated using the PBTK model. Both measured and modeled in vivo EC50s were then correlated with the respective in vitro EC50 values.
structure−activity relationships (QSARs); read-across, grouping or weight-of-evidence methods) that attempted to predict a substance’s toxicological effects based on its physicochemical characteristics or by simply assuming that similar chemical structures result in similar toxicological effects.8−10 To minimize the use of animals, the 3R principle (i.e., reduction, replacement, refinement) requires that animal experiments should be substituted with appropriate alternative test methods whenever possible.11 The practice of using nonexperimental data appears desirable and would allow authorities to regulate chemicals even if data gaps exist. Indeed, predictions regarding toxicological effects using these models are in many cases sufficiently precise.12 However, there are many prominent examples of chemicals with specific modes of action (MOA) for which an accurate prediction regarding toxicity and ecotoxicity would not have been possible at the time that these substances were first regulated, including polychlorinated biphenyls (PCB), bisphenol A (BPA), and alkyl phenols (AP) or tributyltin (TBT), to name only a few. In this context, it could be dangerous to base the risk assessment only on nonexperimental in silico methods.9 In contrast, in vitro bioassays offer a promising experimental alternative to animal studies. Surprisingly, in vitro bioassays played only a minor role in the first registration period under REACH.8,9 Lilienblum et al.12 reviewed the available animal alternatives and concluded that there are already numerous suitable methods for pathway analyses (i.e., for the characterization of the MOAs for certain substances), while the in vitro assessment of more complex processes (e.g., toxicokinetics) continues to require scientific input to meet regulatory requirements. One of the main advantages of in vitro assays is that they can be run under controlled conditions in microplates that allow for high throughput. However, this is a disadvantage with respect to toxicokinetics because the physiology of the organism is not represented. Early in the history of toxicological research on rats (Rattus norvegicus) and mice (Mus musculus), researchers attempted to develop in vitro methods reflecting toxicological effects in vivo. Mason et al.13 were able to show that the median effect concentration EC50 (i.e., the concentration of a chemical of interest that results in a half-maximal effect) for the induction of aryl hydrocarbon hydroxylase (AHH) by polybrominated dibenzo-p-dioxins determined after intraperitoneal (i.p.) injection correlated well with the respective EC50 values measured in the rat hepatoma cell line H4IIE. Similar trends were described in the work of Safe et al.14 for the effects of substituted PCBs, in which in vitro receptor binding affinities positively correlated with the in vivo EC50s for 7-ethoxyresorufin-O-deethylase (EROD) induction. In aquatic ecotoxicology, however, chemicals are most often administered through aqueous exposure medium. For estrogenic activity, a correlation was not observed between in vitro and in vivo data.15 Due to the different physicochemical properties of the substances, the substances can be absorbed at different rates and accumulated to a varying extent in different tissues and organs,16 unlike with i.p. injections. This fact was considered in the early 1990s in the critical body residue (CBR) concept,17 which showed that the chemical concentration in the organism was the central factor for acute toxicity. Hence, the internal concentration at the EC50 level was calculated by multiplying the EC50 with the bioconcentration factor (BCF). A major disadvantage is that the BCF can only be applied under equilibrium conditions.16 Surprisingly, this concept has received less attention in the scientific community for the investigation of
■
MATERIAL AND METHODS Implementation and Extension of the PBTK Model. The multicompartment PBTK model for rainbow trout (Oncorhynchus mykiss) by Nichols et al.,18 with modifications by Stadnicka et al.,19 was reimplemented in Rapid Application Development Studio XE2 (Embarcadero, Langen, Germany). Five different compartments (the liver, richly perfused tissues, poorly perfused tissues, kidney, and fat) as well as the blood were explicitly represented in the model. Each of the compartments is characterized by its volume (fraction of total body weight) and its total lipid and water contents (fraction of tissue wet weight) as well as the blood flow to the compartment (fraction of cardiac output). Model inputs comprise the body wet weight and the total body lipid of the investigated fish and the log Kow and concentration of the test chemical as well as temperature and dissolved oxygen concentration during exposure. Cardiac output, effective respiratory volume, oxygen consumption rate, blood:water, and tissue:blood partitioning coefficients are dynamically calculated from the model inputs. A data set published in the supplement of Stadnicka et al.19 was used to validate the model performance and extended to simulate intraperitoneal (i.p.) injections by adding the appropriate dose of the chemical pollutant to the poorly perfused tissue (PPT) compartment. Furthermore, a simple algorithm for metabolism following Michaelis−Menten kinetics was implemented into the model24 VL
V ·C /P dCL = − max L L dt K m + CL/PL
(1)
where VL is the volume of the liver compartment, CL is the concentration of the chemical in the liver compartment, PL is the partitioning coefficient of the chemical between the blood and liver compartment, Vmax is the maximum velocity of the saturable metabolism, and Km is the Michaelis−Menten constant. Metabolism was only implemented for polycyclic aromatic hydrocarbons (PAHs) because these substances are rapidly metabolized in rainbow trout.25 Values for Km and Vmax were available only for pyrene in rainbow trout. As a first estimate for 3304
dx.doi.org/10.1021/es4053208 | Environ. Sci. Technol. 2014, 48, 3303−3309
Environmental Science & Technology
■
metabolism, these values were applied to all PAHs in the evaluated data set (benzo[k]fluoranthene, benzo[b]fluoranthene, 2,3-benzo[b]fluorene, dibenzo[a,h]anthracene, retene), which is a valid assumption because the substances share common physicochemical properties.26 Vmax was scaled to the respective body weight of the fish by use of eq 2 Vmax = Vmax cbw 0.7
(2)
(top − bottom) 1 + 10(log EC50 − x)·slope
RESULTS AND DISCUSSION
Reimplementation of the PBTK Model. The reimplemented PBTK model was evaluated using a previously published data set.19 The data set consisted of the internal concentrations (Cint) in fish following aqueous exposure to 26 single organic chemicals (39 data points total) and the corresponding exposure conditions (e.g., the concentration of the chemical, temperature, dissolved oxygen concentration, weight of the exposed animal). We then used the PBTK model to predict Cint values under specific exposure conditions and compared those values with the experimentally measured Cint values. The model was able to precisely estimate internal concentrations (Supplementary Figure S1); of all model predictions, 95% deviated less than 10-fold from the measured concentrations, while 82% deviated less than 5-fold. The model performed equally well compared with the previously published implementation by Stadnicka et al.19 In this study, 95% of the model predictions differed less than 10-fold from measured values, while 77% deviated less than 5fold. Correlation of Measured in Vitro and in Vivo Data. To evaluate whether the PBTK model could be used to predict in vivo measures of toxicity (i.e., median effect concentrations (EC50s)) from in vitro data generated using primary hepatocytes, we first compiled a comprehensive data set comprising both experimental in vitro and in vivo data on receptor-mediated effects. We focused on two of the best described and most relevant biomarkers in fish: EROD, an exposure biomarker for dioxins and dioxin-like compounds that is modulated via the cytosolic aryl hydrocarbon receptor (AhR), and Vtg, an exposure biomarker for estrogenic substances under the control of the estrogen receptor (ER). A total of 13 values for in vitro EROD induction were available (Supplementary Table S1), of which eight EC50s for primary hepatocytes were recalculated using relative potency (REP) factors (i.e., the ratio of the EC50 for the well-characterized standard substance 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD) to the chemical of interest, measured in permanent cell lines). Detailed information on the recalculation is given in Supplementary Table S3 and Supplementary Equation (S1). There were 28 in vivo values available, of which 15 were derived from i.p.-injected animals, seven from flow-through exposures and six from semistatic exposures. For Vtg induction, five in vitro and eight in vivo values were available (Supplementary Table S1). For the EROD induction, the in vitro EC50 values were significantly correlated with the experimental in vivo EC50 values when fish were i.p. injected (Figure 1a, R2 = 0.89, p ≤ 0.0001). However, a strong correlation was not found following aqueous exposure (Figure 1b, R2 = 0.31, p = 0.05). In addition, the in vitro EC50s were not significantly correlated with the in vivo data for Vtg induction that comprised only aqueous exposures (Figure 2a, R2 = 0.43, p = 0.08). Our mechanism-specific data were consistent with the conclusion of McCarty and Mackay,17 who showed that the aqueous concentration of a chemical was not a suitable proxy for the internal concentration. This is, however, a common assumption when performing studies with aquatic organisms.33 From the experimental standpoint, the quantification of internal concentrations at the site of toxic action is required to achieve quantitative results in conducting risk assessments that may lead to regulatory decisions.34 Correlation of in Vitro Data with Simulated in Vivo Data. Next, we used the PBTK model to calculate the internal
where Vmaxc is the scaling factor of Vmax, and bw is the body wet weight of the fish.27 Compilation of in Vivo Data Used for Model Evaluation. A comprehensive data set of EROD and Vtg induction in rainbow trout was compiled by performing a literature research using the ISI Web of Knowledge and Science Direct. Furthermore, a review article by Whyte et al.28 on EROD activity in fish was used. Inclusion criteria for in vivo data were as follows: (a) data were derived from aqueous exposure (i.e., flowthrough or semistatic) or i.p. injection; (b) at least three concentration levels were tested in addition to one control treatment; and (c) publications with complete accompanying data (i.e., temperature, weight, age, and sex of the animals) were preferred. If no temperature data were available, 12 °C was used as the default, which is the optimum temperature for rainbow trout.29 The water was assumed to be saturated with oxygen during exposure in all studies, and the dissolved oxygen concentration was calculated according to Weiss.30 Compilation of Bioassay Data Used for Model Evaluation. In vitro data were either directly measured in primary rainbow trout hepatocytes; if no such data were available, the relative potency (REP) values for EROD induction measured in permanent cell lines were used to derive the respective hepatocyte EC50.31,32 Detailed information on the calculation is given in Supplementary Table S3 and Supplementary Equation S1. Calculation of EC50 Values. If not provided in the publications, EC50 values for both in vitro and in vivo experiments were calculated by a four-parameter logistic regression of the log-transformed data in Prism 5 software (GraphPad, San Diego, USA) using eq 3, with top and bottom values set to the maximum and control response, respectively y = bottom +
Article
(3)
where bottom is the y-value of the bottom plateau, top is the yvalue of the top plateau, slope describes the steepness of the curve, and log EC50 is the x-value of the median response. The final data set can be found in the Supporting Information (Supplementary Table S1). Calculation of Internal Concentrations and Correlation Analyses. The PBTK model was used to calculate the internal concentration in the liver compartment (Chep) and whole body (Cint) for in vivo biomarker data at the EC50 level. The toxicodynamics of EROD and Vtg induction were not explicitly represented in the modeling approach. The model outputs are summarized in the Supporting Information (Supplementary Table S2). A correlation analysis of the in vitro data with (a) the experimental in vivo EC50s and (b) the EC50 values based on the modeled Cint and Chep was performed (Spearman rank order correlation) using Prism 5 (GraphPad, San Diego, USA). 3305
dx.doi.org/10.1021/es4053208 | Environ. Sci. Technol. 2014, 48, 3303−3309
Environmental Science & Technology
Article
Figure 1. Correlation between in vitro and in vivo data for EROD induction in rainbow trout based on experimental data. The data were derived either through i.p. injection (a) or flow-through and semistatic aqueous exposure (b). The solid line represents the linear regression line, and dashed lines indicate the 95% confidence interval of the regression. The coefficient of determination (R2) and equation for the regression line using log-transformed data are provided in the graphs.
Figure 2. Correlation between in vitro and in vivo data for Vtg induction in rainbow trout. The correlations were either based on experimental in vivo data (a) or through modeled hepatic in vivo concentrations at the EC50 level (b). The solid line represents the linear regression line, and dashed lines indicate the 95% confidence interval of the regression. The coefficient of determination (R2) and, when applicable, the equation for the regression line using log-transformed data are provided in the graphs.
concentrations in the whole body (Cint) and liver (Chep), the main target organ for both receptor-mediated pathways in fish, at the in vivo EC50 level as reported in the collected literature. Because the model can be used to calculate internal concentrations from both injection and aqueous exposure experiments, it was possible to perform a combined analysis of EROD induction for both dosing methods. We then correlated the experimental in vitro EC50s with the modeled in vivo EC50s based on internal concentrations (Figure 3a-d). We found a strong correlation between the in vitro data and modeled Cint and Chep at the EC50 level (Figure 3a, c). Both Cint and Chep correlated equally well with the reported in vitro EC50s; the coefficient of determination (R2) was 0.83, and 84.4% of all values deviated less than 10-fold from the regression line. Polycyclic aromatic hydrocarbons (PAHs) are easily metabolized in fish livers.25 Thus, we included an algorithm for saturable metabolism according to Michaelis−Menten. The effect on the internal concentration in the whole body Cint was relatively minute; with the modification, 87.5% of the values deviated less than 1 order of magnitude from the regression line (Figure 3b). When focusing on the hepatic concentration Chep (i.e., the organ in which the majority of biotransformation occurs), up to 96.9% of data points deviated by less than 1 order of magnitude (Figure 3d). These findings were applicable to EROD and Vtg induction. We found a significant correlation between the in vitro EC50 for Vtg induction and the modeled Chep at the in vivo EC50 level (Figure 2b, R2 = 0.99, p < 0.0001). PBTK models allow us to predict the concentration of a chemical of interest in virtually any tissue or organ and at any time following exposure, taking the
physiological structure of the organism into account. Even if we did not explicitly account for the potentially large variability in the data collected from the literature, it suggests that even complex and heterogeneous data sets, such as the reactions of different biomarkers in fish tissue, can be scaled from exposure concentrations or doses to internal concentrations, thereby facilitating the potential use of in vitro assays as alternatives to animal studies. In addition, the use of a PBTK model allowed us to compare the results of different exposure routes, which are typically incomparable. For example, injected doses (that are expressed on a per-weight basis) could be compared with aqueous exposure concentrations (that are expressed on a per-volume basis), and dermal and dietary exposures could also be implemented into the model.35,36 Assuming that a submodel describing the toxicokinetics is not necessary for the rapidly induced biomarkers EROD and Vtg, EC50 values based on internal concentrations could even be calculated if fish are exposed to a single concentration level with measurements of the biomarkers performed at different time points, allowing for much faster testing and a reduced reliance on animal experimentation. To bring the proposed modeling methodology to its full potential, suitable and robust in vitro methods are needed to estimate the in vivo hepatic metabolism of a chemical of interest, which was only generally addressed in this study. Several promising and sophisticated methods have been proposed and will need further improvement and development.37,38 Furthermore, it would also be favorable to base in vitro EC50s on the 3306
dx.doi.org/10.1021/es4053208 | Environ. Sci. Technol. 2014, 48, 3303−3309
Environmental Science & Technology
Article
Figure 3. Correlation between experimental in vitro and modeled in vivo data for EROD induction in rainbow trout. Correlations were either based on experimental data derived from i.p. injection (●, (gray)) or flow-through and semistatic aqueous exposure (● (blue)/■ (blue)). The model was used to predict either the internal concentration in the whole body (Cint, a-b) or liver (Chep, c-d). In b and d, saturable hepatic metabolism was implemented in the model for the rapidly metabolized PAHs (■ (blue)). Solid lines represent linear regressions, and dashed lines indicate a 10-fold difference from the regression line. The percentage of values differing from the regression line by less than 10-fold is provided in the lower right corner of the plots. The coefficient of determination (R2) and equation for the regression line using log-transformed data are also provided for reference.
killed carefully prior to liver removal, drastically minimizing suffering compared with lengthy exposure experiments over several days or weeks. To make the presented approach an authentic alternative test method, however, in vitro data should be generated using permanent cell cultures (e.g., the cell line RTL-W1 that was originally derived from rainbow trout liver31,32 and has been used in various studies investigating chemicals,43 soils,44 and sediments45). In addition to lower costs and effort and higher throughput, experiments with such cell lines can be highly standardized and ensure high reproducibility and relatively low interlaboratory variations. Historically, permanent cell lines were cultured in complex media containing fetal calf serum (FCS) or other products derived from animals. Whenever possible, cell cultures should be adapted to serum-free media to further reduce the use of animals.46 Potential Combination of PBTK and Toxicodynamic (TD) Models. The assumption that accounting for toxicodynamics in a separate submodel is not necessary can only be applied if the induction of the biomarker of interest is closely coupled in time with the uptake of the chemical. For EROD and Vtg induction, this assumption appears valid. To investigate biomarkers with slower reaction times (e.g., aneugenic and clastogenic effects, hepatic lesions or carcinogenesis), coupling the PBTK model with toxicodynamic (TD) models is necessary (e.g., Jager et al.;47 PBTK/TD models). At present, this methodology remains rarely used in ecotoxicology and has the potential to provide robust and accurate predictions of the adverse effects from in vitro bioassay data. We conclude that for toxicological and ecotoxicological risk assessments, PBTK models can play an important role in extrapolating the results of in vitro assays to the individual in vivo level and predicting a chemical’s bioaccumulation potential. The presented PBTK model can be easily reparameterised for other
internal concentration in the cell. This however would require extensive chemical analytical work when performing biotests.39 Comparison with Other Methods for in Vitro−in Vivo Extrapolation. Due to the advantages of PBTK models, they are frequently used for IVIVE in pharmacological studies, in which they are typically referred to as physiologically based pharmacokinetic (PBPK) models.20 There are a number of other methods for in vitro−in vivo extrapolation, which often either (a) use simple one-compartment models to forecast the steady-state in vivo concentration40 or (b) simply determine equipotent doses and concentrations in the in vivo and in vitro studies, respectively (also called the benchmark dose approach).41 Compared with these simple approaches, the data requirements for the parametrization of PBTK models for a new species are relatively extensive. If the model already exists, however, only information on log Kow and biotransformation (when necessary) is required. The methodology proposed in this study has a number of major advantages: it is (a) a kinetic model that can predict the internal concentrations at any time following exposure; (b) a multicompartment model that comprises many different tissues and organs and can be extended to explicitly represent other organs of interest if desired, such as the brain;42 and (c) a mechanistic model that can offer deeper insights into the physiology behind any observed pattern or phenomenon. Once established, the model can be used to evaluate virtually any end point or biomarker; and PBTK models for different fish species could be used for probabilistic risk assessment of bioaccumulation and toxic effects, which is of particular interest in the context of the REACH regulation. Suitability of in Vitro Assays for the Aquatic Risk Assessment of Chemicals. Primary hepatocytes must be obtained from live fish, making them a questionable alternative to animal experiments. However, animals are anesthetised and 3307
dx.doi.org/10.1021/es4053208 | Environ. Sci. Technol. 2014, 48, 3303−3309
Environmental Science & Technology
■
fish species. PBTK models are already available for many fish species, including dogfish sharks,48 brook trout,49 lake trout,50 channel catfish,51 fathead minnows, Japanese medaka,52 and tilapia.53 Thus, the relevance of this study goes far beyond toxicological studies with rainbow trout. The results can easily be applied to other aquatic species or environments by establishing comparable relationships between the aforementioned PBTK models and toxicological studies. In this way, the proposed methodology has the potential to facilitate a probabilistic aquatic risk assessment of industrial chemicals. Recently, the development of a PBTK model for zebrafish (Danio rerio) supported this perspective because it is one of the most widely used species in (eco)toxicological studies.54 In addition, the combination of in vitro tests to investigate the effects and metabolic transformation of chemical pollutants in combination with PBTK models can allow for a deeper mechanistic understanding of the differences in species sensitivity due to differences in receptors,15 physiological structure,19 or metabolic capacity. The present study demonstrates that PBTK models can be used to ordinate important biochemical markers of mechanismspecific effects (i.e., EROD and Vtg) to correlate with experimental in vitro data. In this context, implementing toxicodynamics into the model was unnecessary because the biochemical markers of interest generally reacted rapidly during contaminant exposure. If more complex effects (e.g., aneugenic or clastogenic effects, hepatic lesions or carcinogenesis) need to be predicted, specific TD submodels for the estimation of temporal dynamics must be developed. From a regulator’s standpoint, we believe that nonexperimental in silico methods for in vitro−in vivo extrapolation have the potential to achieve an overall reduction in the number of toxicological and ecotoxicological experiments with live animals when combined with experimental in vitro test methods while maintaining an equal level of protection.
■
ASSOCIATED CONTENT
Detailed information on the experimental in vitro and in vivo data and performance of the reimplemented model as well as calculation of in vitro hepatocyte EC50s from REP values derived from permanent cell lines. This material is available free of charge via the Internet at http://pubs.acs.org.
AUTHOR INFORMATION
Corresponding Author
*Phone: 49 (0)241 − 80/26686. Fax: 49 (0)241 − 80/22182. Email:
[email protected]. Notes
The authors declare no competing financial interest.
■
REFERENCES
(1) Crutzen, P. J. Geology of mankind. Nature 2002, 415 (6867), 23− 23. (2) Rockstrom, J.; Steffen, W.; Noone, K.; Persson, A.; Chapin, F. S.; Lambin, E. F.; Lenton, T. M.; Scheffer, M.; Folke, C.; Schellnhuber, H. J.; Nykvist, B.; de Wit, C. A.; Hughes, T.; van der Leeuw, S.; Rodhe, H.; Sorlin, S.; Snyder, P. K.; Costanza, R.; Svedin, U.; Falkenmark, M.; Karlberg, L.; Corell, R. W.; Fabry, V. J.; Hansen, J.; Walker, B.; Liverman, D.; Richardson, K.; Crutzen, P.; Foley, J. A. A safe operating space for humanity. Nature 2009, 461 (7263), 472−475. (3) Schwarzman, M. R.; Wilson, M. P. New science for chemicals policy. Science 2009, 326 (5956), 1065−1066. (4) EC, Regulation no. 1907/2006 of the European Parliament and of the Council concerning the registration, evaluation, authorization and restriction of chemicals. 2006. (5) Schwarzenbach, R. P.; Escher, B. I.; Fenner, K.; Hofstetter, T. B.; Johnson, C. A.; von Gunten, U.; Wehrli, B. The challenge of micropollutants in aquatic systems. Science 2006, 313 (5790), 1072− 1077. (6) Hartung, T.; Rovida, C. Chemical regulators have overreached. Nature 2009, 460 (7259), 1080−1081. (7) Schulte, C.; Tietjen, L.; Bambauer, A.; Fleischer, A. Five years REACH - lessons learned and first experiences. I. An authorities’ view. Environ. Sci. Eur. 2012, 24 (1), 31. (8) ECHA, The use of alternatives to testing on animals for the REACH regulation. European Chemicals Agency, Helsinki, Finland, 2011. (9) Spielmann, H.; Sauer, U. G.; Mekenyan, O. A critical evaluation of the 2011 ECHA reports on compliance with the REACH and CLP regulations and on the use of alternatives to testing on animals for compliance with the REACH regulation. Altern. Lab. Anim. 2011, 39, 481−93. (10) Gilbert, N. Data gaps threaten chemical safety law. Nature 2011, 475, 150−151. (11) Russell, W. M. S.; Burch, R. L. The principles of humane experimental technique; special ed., reprinted 1992; UFAW: London, 1959. (12) Lilienblum, W.; Dekant, W.; Foth, H.; Gebel, T.; Hengstler, J. G.; Kahl, R.; Kramer, P. J.; Schweinfurth, H.; Wollin, K. M. Alternative methods to safety studies in experimental animals: role in the risk assessment of chemicals under the new european chemicals legislation (REACH). Arch. Toxicol. 2008, 82 (4), 211−236. (13) Mason, G.; Zacharewski, T.; Denomme, M. A.; Safe, L.; Safe, S. Polybrominated dibenzo-p-dioxins and related compounds: quantitative in vivo and in vitro structure-activity relationships. Toxicology 1987, 44, 245−255. (14) Safe, S.; Bandiera, S.; Sawyer, T.; Zmudzka, B.; Mason, G.; Romkes, M.; Denomme, M. A.; Sparling, J.; Okey, A. B.; Fujita, T. Effects of structure on binding to the 2,3,7,8-TCDD receptor protein and AHH induction - halogenated biphenyls. Environ. Health Perspect. 1985, 61 (SEP), 21−33. (15) Segner, H.; Navas, J. M.; Schäfers, C.; Wenzel, A. Potencies of estrogenic compounds in in vitro screening assays and in life cycle tests with zebrafish in vivo. Ecotox. Environ. Saf. 2003, 54 (3), 315−322. (16) Mackay, D.; Fraser, A. Bioaccumulation of persistent organic chemicals: mechanisms and models. Environ. Pollut. 2000, 110 (3), 375−391. (17) McCarty, L. S.; Mackay, D. Enhancing ecotoxicological modeling and assessment. Body residues and modes of toxic action. Environ. Sci. Technol. 1993, 27 (9), 1718−1728. (18) Nichols, J. W.; McKim, J. M.; Lien, G. J.; Hoffman, A. D.; Bertelsen, S. L. Physiologically based toxicokinetic modeling of three waterborne chloroethanes in rainbow trout (Oncorhynchus mykiss). Toxicol. Appl. Pharmacol. 1991, 110 (3), 374−389. (19) Stadnicka, J.; Schirmer, K.; Ashauer, R. Predicting concentrations of organic chemicals in fish by using toxicokinetic models. Environ. Sci. Technol. 2012, 46 (6), 3273−3280. (20) Yoon, M.; Campbell, J.; Andersen, M.; Clewell, H. Quantitative in vitro to in vivo extrapolation of cell-based toxicity assay results. Crit. Rev. Toxicol. 2012, 42 (8), 633−652.
S Supporting Information *
■
Article
ACKNOWLEDGMENTS
The study was performed as part of the project “DioRAMA Assessment of the dioxin-like activity in sediments and fish for sediment evaluation” and received funds from the German Federal Ministry of Transport, Building and Urban Development. M.B. received a personal stipend from the German National Academic Foundation (“Studienstiftung des deutschen Volkes”). T.P. and T.B. were funded by the German Federal Environment Agency (“Further development of criteria for bioaccumulation under REACH”, FKZ 3711 63405 2). 3308
dx.doi.org/10.1021/es4053208 | Environ. Sci. Technol. 2014, 48, 3303−3309
Environmental Science & Technology
Article
using measured rates of in vitro intrinsic clearance. Environ. Toxicol. Chem. 2013, 32 (7), 1611−1622. (39) Gülden, M.; Mörchel, S.; Seibert, H. Factors influencing nominal effective concentrations of chemical compounds in vitro: cell concentration. Toxicol. In Vitro 2001, 15 (3), 233−243. (40) Rotroff, D. M.; Wetmore, B. A.; Dix, D. J.; Ferguson, S. S.; Clewell, H. J.; Houck, K. A.; LeCluyse, E. L.; Andersen, M. E.; Judson, R. S.; Smith, C. M.; Sochaski, M. A.; Kavlock, R. J.; Boellmann, F.; Martin, M. T.; Reif, D. M.; Wambaugh, J. F.; Thomas, R. S. Incorporating human dosimetry and exposure into high-throughput in vitro toxicity screening. Toxicol. Sci. 2010, 117 (2), 348−358. (41) Piersma, A. H.; Janer, G.; Wolterink, G.; Bessems, J. G. M.; Hakkert, B. C.; Slob, W. Quantitative extrapolation of in vitro whole embryo culture embryotoxicity data to developmental toxicity in vivo using the benchmark dose approach. Toxicol. Sci. 2008, 101 (1), 91− 100. (42) Abbas, R.; Hayton, W. A physiologically based pharmacokinetic and pharmacodynamic model for paraoxon in rainbow trout. Toxicol. Appl. Pharmacol. 1997, 145, 192−201. (43) Olsman, H.; Engwall, M.; Kammann, U.; Klempt, M.; Otte, J.; van Bavel, B.; Hollert, H. Relative differences in aryl hydrocarbon receptormediated response for 18 polybrominated and mixed halogenated dibenzo-p-dioxins and -furans in cell lines from four different species. Environ. Toxicol. Chem. 2007, 26 (11), 2448−2454. (44) Andersson, E.; Rotander, A.; Kronhelm, T.; Berggren, A.; Ivarsson, P.; Hollert, H.; Engwall, M. AhR agonist and genotoxicant bioavailability in a PAH-contaminated soil undergoing biological treatment. Environ. Sci. Pollut. Res. 2009, 16 (5), 521−530. (45) Wölz, J.; Brack, W.; Moehlenkamp, C.; Claus, E.; Braunbeck, T.; Hollert, H. Effect-directed analysis of Ah receptor-mediated activities caused by PAHs in suspended particulate matter sampled in flood events. Sci. Total Environ. 2010, 408 (16), 3327−3333. (46) Gstraunthaler, G. Alternatives to the use of fetal bovine serum: serum-free cell culture. ALTEX 2003, 20, 275−281. (47) Jager, T.; Albert, C.; Preuss, T. G.; Ashauer, R. General unified threshold model of survival - a toxicokinetic-toxicodynamic framework for ecotoxicology. Environ. Sci. Technol. 2011, 45 (7), 2529−2540. (48) Bungay, P.; Dedrick, R.; Guarino, A. Pharmacokinetic modeling of the dogfish shark (Squalus acanthias): Distribution and urinary and biliary excretion of phenol red and its glucuronide. J. Pharmacokinet. Biopharm. 1976, 4 (5), 377−388. (49) Nichols, J. W.; Jensen, K. M.; Tietge, J. E.; Johnson, R. D. Physiologically based toxicokinetic model for maternal transfer of 2,3,7,8-tetrachlorodibenzo-p-dioxin in brook trout (Salvelinus fontinalis). Environ. Toxicol. Chem. 1998, 17 (12), 2422−2434. (50) Lien, G. J.; McKim, J. M.; Hoffman, A. D.; Jenson, C. T. A physiologically based toxicokinetic model for lake trout (Salvelinus namaycush). Aquat. Toxicol. 2001, 51 (3), 335−350. (51) Nichols, J. W.; McKim, J. M.; Lien, G. J.; Hoffman, A. D.; Bertelsen, S. L.; Gallinat, C. A. Physiologically-based toxicokinetic modeling of three waterborne chloroethanes in channel catfish Ictalurus punctatus. Aquat. Toxicol. 1993, 27 (1−2), 83−111. (52) Lien, G. J.; McKim, J. M. Predicting branchial and cutaneous uptake of 2,2′,5,5′-tetrachlorobiphenyl in fathead minnows (Pimephales promelas) and Japanese medaka (Oryzias latipes): rate limiting factors. Aquat. Toxicol. 1993, 27 (1−2), 15−31. (53) Liao, C.-M.; Liang, H.-M.; Chen, B.-C.; Singh, S.; Tsai, J.-W.; Chou, Y.-H.; Lin, W.-T. Dynamical coupling of PBPK/PD and AUCbased toxicity models for arsenic in tilapia Oreochromis mossambicus from blackfoot disease area in Taiwan. Environ. Pollut. 2005, 135 (2), 221−233. (54) Péry, A. R. R.; Devillers, J.; Brochot, C.; Mombelli, E.; Palluel, O.; Piccini, B.; Brion, F.; Beaudouin, R. A physiologically based toxicokinetic model for the zebrafish Danio rerio. Environ. Sci. Technol. 2013, 48 (1), 781−790.
(21) Louisse, J.; Verwei, M.; Woutersen, R. A.; Blaauboer, B. J.; Rietjens, I. M. Toward in vitro biomarkers for developmental toxicity and their extrapolation to the in vivo situation. Expert Opin. Drug Metab. Toxicol. 2012, 8 (1), 11−27. (22) Hecker, M.; Hollert, H. Endocrine disruptor screening: regulatory perspectives and needs. Environ. Sci. Eur. 2011, 23 (1), 15. (23) Weber, R.; Gaus, C.; Tysklind, M.; Johnston, P.; Forter, M.; Hollert, H.; Heinisch, E.; Holoubek, I.; Lloyd-Smith, M.; Masunaga, S.; Moccarelli, P.; Santillo, D.; Seike, N.; Symons, R.; Torres, J.; Verta, M.; Varbelow, G.; Vijgen, J.; Watson, A.; Costner, P.; Woelz, J.; Wycisk, P.; Zennegg, M. Dioxin- and POP-contaminated sitescontemporary and future relevance and challenges. Environ. Sci. Pollut. Res. 2008, 15 (5), 363−393. (24) Law, F. C. P.; Abedini, S.; Kennedy, C. J. A biologically based toxicokinetic model for pyrene in rainbow trout. Toxicol. Appl. Pharmacol. 1991, 110 (3), 390−402. (25) Rand, G. M.; Petrocelli, S. R. Fundamentals Of Aquatic Toxicology: Methods And Applications; Hemisphere Publ. Corp.: New York, 1985; pp 385−409. (26) de Maagd, P. G.-J.; Verthaak, A. D. Biotransformation of PAHs and their carcinogenic effects in fish. In PAHs and Related Compounds: Biology; Neilson, A. H., Ed.; Springer: Berlin, Heidelberg, NY, 1993. (27) Travis, C. C. Interspecies and dose-route extrapolations. In Drinking Water and Health. Pharmacokinetics in Risk Assessment; National Academy Press: Washington, DC, 1987; Vol. 8. (28) Whyte, J. J.; Jung, R. E.; Schmitt, C. J.; Tillitt, D. E. Ethoxyresorufin-O-deethylase (EROD) activity in fish as a biomarker of chemical exposure. CRC Crit. Rev. Toxicol. 2000, 30 (4), 347−570. (29) Johnson, J. E.; Kramer, R. P.; Larson, E.; Bonebrake, B. L. Trout growth, harvest, survival and microhabitat selection in the Green River, Utah, 1978−1982. Final Report, Flaming Gorge Tailwater Fisheries Investigations; Utah Division of Wildlife Resources: Salt Lake City, 1987. (30) Weiss, R. F. The solubility of nitrogen, oxygen and argon in water and seawater. Deep-Sea Res. Oceanogr. Abstr. 1970, 17 (4), 721−735. (31) Bols, N. C.; Schirmer, K.; Joyce, E. M.; Dixon, D. G.; Greenberg, B. M.; Whyte, J. J. Ability of polycyclic aromatic hydrocarbons to induce 7-ethoxyresorufin-O-deethylase activity in a trout liver cell line. Ecotoxicol. Environ. Saf. 1999, 44 (1), 118−128. (32) Lee, L. J.; Clemons, J.; Bechtel, D.; Caldwell, S.; Han, K.-B.; Pasitschniak-Arts, M.; Mosser, D.; Bols, N. Development and characterization of a rainbow trout liver cell line expressing cytochrome P450-dependent monooxygenase activity. Cell Biol. Toxicol. 1993, 9 (3), 279−294. (33) Barron, M. G.; Anderson, M. J.; Lipton, J.; Dixon, D. G. Evaluation of critical body residue QSARS for predicting organic chemical toxicity to aquatic organisms. SAR QSAR Environ. Res. 1997, 6 (1−2), 47−62. (34) Solomon, K. R.; Brock, T. C.; De Zwart, D.; Dyer, S. D.; Posthuma, L.; Richards, S.; Sanderson, H.; Sibley, P.; van den Brink, P. J. Extrapolation Practice for Ecotoxicological Effect Characterization of Chemicals; CRC Press: 2008. (35) Nichols, J. W.; Fitzsimmons, P. N.; Whiteman, F. W.; Dawson, T. D.; Babeu, L.; Juenemann, J. A physiologically based toxicokinetic model for dietary uptake of hydrophobic organic compounds by fish: I. feeding studies with 2,2′,5,5′-tetrachlorobiphenyl. Toxicol. Sci. 2004, 77 (2), 206−218. (36) Nichols, J. W.; Mckim, J. M.; Lien, G. J.; Hoffman, A. D.; Bertelsen, S. L.; Elonen, C. M. A physiologically based toxicokinetic model for dermal absorption of organic chemicals by fish. Toxicol. Sci. 1996, 31 (2), 229−242. (37) Escher, B. I.; Cowan-Ellsberry, C. E.; Dyer, S.; Embry, M. R.; Erhardt, S.; Halder, M.; Kwon, J.-H.; Johanning, K.; Oosterwijk, M. T. T.; Rutishauser, S.; Segner, H.; Nichols, J. Protein and lipid binding parameters in rainbow trout (Oncorhynchus mykiss) blood and liver fractions to extrapolate from an in vitro metabolic degradation assay to in vivo bioaccumulation potential of hydrophobic organic chemicals. Chem. Res. Toxicol. 2011, 24 (7), 1134−1143. (38) Nichols, J. W.; Huggett, D. B.; Arnot, J. A.; Fitzsimmons, P. N.; Cowan-Ellsberry, C. E. Toward improved models for predicting bioconcentration of well-metabolized compounds by rainbow trout 3309
dx.doi.org/10.1021/es4053208 | Environ. Sci. Technol. 2014, 48, 3303−3309
